Cloud computing offers a variety of services, including the dynamic availability of computing resources. Its infrastructure is designed to support the accessibility and availability of various consumer services via the Internet. The number of data centers allow the allocation of the applications, and the process of data in the cloud is increasing over time. This implies high energy consumption, thus contributing to large emissions of CO2 gas. For this reason, solutions are needed to minimize this power consumption, such as virtualization, migration, consolidation, and efficient traffic-aware virtual machine scheduling. In this article, the authors propose two efficient strategies for VM scheduling. SchedCT approach is based on dynamic CPU utilization and temperature thresholds. SchedCR approach takes into consideration dynamic CPU utilization, RAM capacity, and temperature thresholds. These approaches have efficiently decreased the energy consumption of the data centers, the number of VM migrations, and SLA violations, and this reduces, therefore, the emission of CO2 gas.
CPU utilization prediction method based on composite model
